Python Cloud functions throws error intermittently for requests > 10 Kb [149753461]

Can't Repro

Bug

Status Update

No update yet.

Description

so...@google.com

created issue #1

Feb 18, 2020 06:06PM

Observed behavior:
When sending an HTTP Post text payload over 10KB cloud functions reject it

Expected behavior: Payload size limit is over 10 mb according to this doc [1] so request over 10 kb and below 10 mb should be processed

Steps to reproduce:

1. Follow the hello world quickstart (in python) [2]
2 Then, you need to send a post request as follows:

requests.post('api/hello_http',data={'data':data}), where the data variable contains a text payload larger than 15KB.
3. This will issue the error message 'Error: could not handle the request'.
4. If you switch this command to

requests.post('api/hello_http', json={'data':data}), everything works fine.

[1]

https://cloud.google.com/functions/quotas#resource_limits
[2]

https://cloud.google.com/functions/docs/first-python

Comments

so...@google.com <so...@google.com> #2Feb 18, 2020 06:09PM

An internal bug has been created for this, additional feedback will be provided through here

ed...@google.com <ed...@google.com> #3Apr 3, 2020 12:29AM

Hi Team, is there any updates on how is this issue being addressed?

ti...@google.com <ti...@google.com> #4Oct 23, 2020 04:35AM

Can you create a full sample for reproducing this issue? That would make it easier to see this issue and look into the error.

br...@google.com <br...@google.com> #5Feb 11, 2021 06:56PM

I have a customer asking how to resolve this as well, any updates?

m....@combocurve.com <m....@combocurve.com> #6May 17, 2021 11:16PM

We are experiencing a possibly similar issue on Python CFs starting May 3rd:

Function execution took 342 ms, finished with status: 'connection error'

This SO Answer sent me here: https://stackoverflow.com/a/60714504/5003820

Does it seem to be the same issue?

na...@gmail.com <na...@gmail.com> #7Jun 8, 2021 02:21PM

We are also facing this issue very intermittently but are causing problem at our client side. How can we resolve them

ha...@gmail.com <ha...@gmail.com> #8Jun 25, 2021 04:54PM

Is there any update on this?
I'm getting random connection errors even when the size of the request is less than 1 kb.
The function is being called every hour using Cloud Scheduler and the failures don't happen at any particular time in the day, it's completely random.

ca...@google.com <ca...@google.com> #9Aug 25, 2021 01:19AM

Hi,

the product team continues working on this.

fr...@gmail.com <fr...@gmail.com> #10Nov 19, 2021 01:12PM

We are having the same issue with PHP function.

pa...@google.com <pa...@google.com> #11Dec 2, 2021 04:12PM

Hi.

The Engineering team is still working on this issue, please expect updates on this channel.

je...@twinkl.com.au <je...@twinkl.com.au> #12Dec 2, 2021 04:38PM

I see this is still open, today we attempted to develop a PHP74 cloud function to replace our existing NodeJS cloud function, as an event handler for SendGrid event notification webhooks.
Unfortunately, we are seeing intermittent blank body content (headers intact) - for the problematic requests, the content-length header is missing and chunked transfer encoding appears to be in play.
The specification of the SG Webhook states that calls will be made every 30s OR when content reaches 768kb (meaning there should be no 10MB cloud function limit in play here).

There appears to be a number of historical bugs with FASTCGI and chunked encoding, plus some with ngnix, both of which appear to be in play looking at the cloud function startup logs. It looks like the only option is to revert to NodeJS whilst this issue is resolved... it'd be interesting to know whether the issue we are seeing is related to this same issue, or whether we ought to create a new one.

gu...@google.com <gu...@google.com> #13Dec 3, 2021 06:24PM

Hi.

The Engineering team is still working on this issue, please expect updates on this channel.

al...@gmail.com <al...@gmail.com> #14Feb 20, 2022 11:35PM

So, no news?

ga...@google.com <ga...@google.com> #15Feb 21, 2022 10:39PM

Hi,

product team is still working on this issue, please expect updates on this channel.

Thanks.

go...@google.com <go...@google.com> #16Apr 29, 2022 03:44PM

Hi,

I have a customer facing the similar issue. go/sf/29759975, any updates on this?

ta...@retail-ai.jp <ta...@retail-ai.jp> #17Jul 8, 2022 02:27AM

Hi,

How long google engineers are going to work on this? It's been 2 years ...... Should we pay extra pennies for the escalation :)

Message last modified on Jul 8, 2022 02:30AM

lu...@dim28.ch <lu...@dim28.ch> #18Jul 18, 2022 07:12PM

I am also affected by this. Please!

va...@google.com <va...@google.com> Nov 2, 2023 07:12AM

Assigned to su...@google.com.

su...@google.com <su...@google.com> Nov 13, 2023 07:29AM

Reassigned to gc...@google.com.

an...@trendmicro.com <an...@trendmicro.com> #19Jan 9, 2024 02:37AM

Is there any update on this case?

ja...@gmail.com <ja...@gmail.com> #20Jan 9, 2024 12:16PM

It's really a shame that the platform ignored the correction of such an error for so many years. Really pitiful

ba...@google.com <ba...@google.com> Feb 29, 2024 08:57AM

Reassigned to su...@google.com.

su...@google.com <su...@google.com> #21Mar 5, 2024 12:39PM

Status: Won't Fix (Not Reproducible)

Hello,

Thank you for your cooperation.

We have an update from the engineering team.Engineering team is no longer able to reproduce this. It was tested for python 3.8 and works as expected. Hence it might be fixed for you as well.. I am going to close this thread which will no longer be monitored.If you have any further questions or need assistance with any other matter, please don't hesitate to create a new issue on the Issue Tracker. We are here to help.